Peak Integration: Where Two Analysts Disagree
Two analysts given the same chromatogram will not always produce the same purity figure. The trace is fixed; the areas are not. Integration is the step where judgement enters an otherwise mechanical measurement, and it is invisible on a certificate.
Where the baseline goes
Every peak area is measured relative to a baseline the software draws underneath it. On a flat, quiet trace this is uncontroversial. In gradient elution the baseline rises as organic content increases, and the software must decide how to project it under each peak.
A baseline drawn slightly high reduces the measured area; slightly low increases it. For a large, well-resolved main peak the effect is minor. For small impurity peaks near the reporting threshold it is decisive, and it determines whether they are counted at all. Limit of detection and limit of quantitation covers the thresholds involved.
Valley-to-valley versus drop-line
When two peaks are partially resolved, the area has to be divided between them. Two conventions are in common use and they give different answers.
- Drop-line — a vertical line is dropped from the valley between the peaks to the baseline, splitting the overlap down the middle. Simple, and it systematically assigns some of each peak to the other.
- Valley-to-valley — the baseline is drawn between valleys, treating each peak as sitting on the tail of its neighbour. This tends to reduce the area assigned to the smaller peak.
Neither is correct in an absolute sense. Both are conventions, and a method should state which it uses. Where a shoulder sits on a main peak, the choice can move a reported purity by a meaningful fraction of a percent.
Shoulders, riders and unresolved species
A shoulder is the visible edge of an unresolved impurity. Whether it is integrated as a separate peak or absorbed into the main one is an operator decision, and absorbing it always raises purity. This is not usually dishonesty; software defaults frequently make the choice without anyone reviewing it.
The related case is a peak that is not resolved enough to produce a shoulder at all, which is counted as target by every convention. Co-elution covers the detection methods for that.
Why this bounds the precision of a purity figure
Integration variability is one of the largest contributors to the uncertainty on a reported purity, and it is why treating 98.4% and 98.7% as different numbers is usually an error. Measurement uncertainty works through the arithmetic.
It is also the strongest argument for publishing the chromatogram rather than the percentage. A number cannot be re-integrated. A trace can, by anyone who doubts the result.
What it does not affect
Integration changes proportions, not identity. It has no bearing on whether the measured mass matches the calculated mass for the claimed sequence, and none at all on sterility or endotoxin, which are separate tests outside the scope of any chromatographic method. Endotoxin, sterility and the limits of research grade covers that boundary.
All material is supplied for laboratory research use only. It is not a drug, not a supplement, and not for use in humans or animals.
